Zhenfeng Zhu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Zhenfeng Zhu has a total of 188 co-authors. Here are the top co-authors by number of shared publications:
- Yao Zhao
Institute of Information Science, Beijing Jiaotong University, Beijing, China and Visual Intelligence + X International Joint Laboratory, Beijing, China
57 shared publications - Shuai Zheng
Beijing Jiaotong University
24 shared publications Yao Zhao
18 shared publications- Youru Li
College of Computer Science, Beijing University of Technology, Beijing, China
17 shared publications - Zhizhe Liu
School of Information and Communication Engineering, Beijing Information Science and Technology University, Beijing, China
14 shared publications - Xingxing Zhang
School of Computer Science, Tsinghua University, Beijing, China
13 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (11) , 14105-14120
Aligning Condensed Graph via Hashing: A New Insight for Federated Graph Learning
Federated Graph Learning (FGL) aims to maximize the benefits of each graph owner, which is a common form of distributed graph learning under privacy-preserving conditions. As the landscape of local clients becomes increasingly diverse in terms of both model architectures and topological complexities, graph heterogeneity turns out to be one of the significant challenges to efficient collaboration among clients. Beyond existing paradigms, we delve a fresh insight into revisiting FGL as a semantic condensed graph alignment problem in this work. From this perspective, HashFGL is proposed for heterogeneous FGL through aligning condensed graphs via hashing in a symbiotic space. Specifically, the core of HashFGL lies in that it introduces a cross-client symbiotic space to facilitate effective collaboration. Within this space, an efficient hash-based semantic encoding strategy is proposed to model each local client while balancing coordinated resilience and semantic consistency. Furthermore, we derive an elaborate graph condenser based on the above strategy, which condenses original graphs with semantics and structure-preserving property, to maintain the effectiveness of condensed graph alignment for FGL. Formal theoretical analysis further reveals that HashFGL can effectively alleviate the problem of graph heterogeneity. Experimental results on three large-scale graphs, employing standard partitioning strategies and a pioneering, more realistic partitioning that we introduced, demonstrate the efficacy and scalability of HashFGL.
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (11) , 14105-14120
Aligning Condensed Graph via Hashing: A New Insight for Federated Graph Learning
Federated Graph Learning (FGL) aims to maximize the benefits of each graph owner, which is a common form of distributed graph learning under privacy-preserving conditions. As the landscape of local clients becomes increasingly diverse in terms of both model architectures and topological complexities, graph heterogeneity turns out to be one of the significant challenges to efficient collaboration among clients. Beyond existing paradigms, we delve a fresh insight into revisiting FGL as a semantic condensed graph alignment problem in this work. From this perspective, HashFGL is proposed for heterogeneous FGL through aligning condensed graphs via hashing in a symbiotic space. Specifically, the core of HashFGL lies in that it introduces a cross-client symbiotic space to facilitate effective collaboration. Within this space, an efficient hash-based semantic encoding strategy is proposed to model each local client while balancing coordinated resilience and semantic consistency. Furthermore, we derive an elaborate graph condenser based on the above strategy, which condenses original graphs with semantics and structure-preserving property, to maintain the effectiveness of condensed graph alignment for FGL. Formal theoretical analysis further reveals that HashFGL can effectively alleviate the problem of graph heterogeneity. Experimental results on three large-scale graphs, employing standard partitioning strategies and a pioneering, more realistic partitioning that we introduced, demonstrate the efficacy and scalability of HashFGL.
- Open Access
- 21 October 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Information Systems
- Vol. 44 (8) , 1-33
Information-Bottlenecked Variational Autoencoder for Top- N Recommendation via Gating Mechanism
Generative models for top- \( N \) recommendation have garnered significant attention, with Variational Autoencoder (VAE) emerging as a promising approach for modeling user preferences. Yet, traditional VAE-based models encounter two major challenges: simplistic priors may cause posterior collapse, resulting in ineffective latent structures; meanwhile, limited decoder capabilities hinder the processing of complex data and the extraction of task-relevant information, potentially affecting generative performance. To address these issues, we propose VAEinfox , short for Information-Bottlenecked Variational Autoencoder via Gating Mechanism, a simple yet effective framework that mitigates posterior collapse in VAE. VAEinfox introduces an Adaptive Compressed Representation (ACR) module based on the Information Bottleneck (IB) principle to efficiently and adaptively compress and select latent variables, mitigating posterior collapse. It leverages the Relaxed Bernoulli distribution for differentiable compressed representations and employs a dynamic compression probability selection mechanism with a trainable soft mask activation function to automatically adjust latent variable sparsity during training. Then, a joint training objective maximizes mutual information using tshe Jensen–Shannon estimator to minimize latent redundancy. In particular, the generative module in VAEinfox incorporates a Mixture-of-Experts (MoE) paradigm, enhancing representation learning and the decoder’s ability to process complex data and extract task-relevant information. These dual enhancements improve the model’s generative performance while also enhancing its universality and efficiency, particularly when handling implicit feedback. Extensive experiments on eight real-world datasets demonstrate that the proposed framework achieves better results than existing competitors and tackles the specified challenges in top- \( N \) recommendation tasks.- 21 October 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Information Systems
- Vol. 44 (8) , 1-33
Open AccessInformation-Bottlenecked Variational Autoencoder for Top- N Recommendation via Gating Mechanism
Generative models for top- \( N \) recommendation have garnered significant attention, with Variational Autoencoder (VAE) emerging as a promising approach for modeling user preferences. Yet, traditional VAE-based models encounter two major challenges: simplistic priors may cause posterior collapse, resulting in ineffective latent structures; meanwhile, limited decoder capabilities hinder the processing of complex data and the extraction of task-relevant information, potentially affecting generative performance. To address these issues, we propose VAEinfox , short for Information-Bottlenecked Variational Autoencoder via Gating Mechanism, a simple yet effective framework that mitigates posterior collapse in VAE. VAEinfox introduces an Adaptive Compressed Representation (ACR) module based on the Information Bottleneck (IB) principle to efficiently and adaptively compress and select latent variables, mitigating posterior collapse. It leverages the Relaxed Bernoulli distribution for differentiable compressed representations and employs a dynamic compression probability selection mechanism with a trainable soft mask activation function to automatically adjust latent variable sparsity during training. Then, a joint training objective maximizes mutual information using tshe Jensen–Shannon estimator to minimize latent redundancy. In particular, the generative module in VAEinfox incorporates a Mixture-of-Experts (MoE) paradigm, enhancing representation learning and the decoder’s ability to process complex data and extract task-relevant information. These dual enhancements improve the model’s generative performance while also enhancing its universality and efficiency, particularly when handling implicit feedback. Extensive experiments on eight real-world datasets demonstrate that the proposed framework achieves better results than existing competitors and tackles the specified challenges in top- \( N \) recommendation tasks. - Open Access
- 5 September 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 20 (8) , 1-32
Trust-Aware Enterprise Credit Risk Prediction via Variational Autoencoder in Supply Chain Finance
The growing complexity of transaction networks, dynamic enterprise interactions, and rising financial uncertainties in modern supply chains have made accurate credit risk assessment for small and medium enterprises (SMEs) in supply chain finance (SCF) a critical priority. However, traditional credit prediction models struggle to capture meaningful latent risk patterns and dynamic risk propagation within supply chain networks, limiting their effectiveness, robustness, and interpretability in supporting trustworthy decision-making in SCF ecosystems. To this end, we propose TrustVAE , a novel end-to-end generative paradigm that jointly models iterative node aggregation, trust-aware message passing, and spatio-temporal dynamics for credit risk assessment in SCF. Specifically, TrustVAE leverages the Iterative Latent Risk Aggregation (ILRA) module to extract latent risk representations from raw enterprise data in an unsupervised manner through deep nonlinear transformations and iterative clustering, capturing inherent structural properties that extend beyond the static financial indicators used in conventional models. Meanwhile, the Bounded Graph Attention Network (BGAN) module integrates trust-aware bounded confidence into graph-based risk propagation, enhancing the fidelity of diffusion modeling by suppressing risk transmission across low-trust enterprise edges. Furthermore, TrustVAE introduces the Temporal Graph Autoregressive Flow (TGAF) module, which couples temporal dependencies with autoregressive flow modeling to characterize credit risk evolution across multi-scale temporal horizons (e.g., daily, weekly, and monthly), jointly capturing dynamic patterns in both temporal and graph-structured domains. Extensive experiments on real-world datasets demonstrate that TrustVAE outperforms existing competitors, achieving average absolute improvements of 1.59%, 3.17%, 1.93%, and 2.11% in ROC-AUC, AR, Accuracy, and F1 Score, respectively. The unified design of TrustVAE strengthens the practical foundation of information systems in addressing credit risk within data-intensive, networked environments, offering novel insights into generative modeling for credit risk prediction.
- 5 September 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 20 (8) , 1-32
Open AccessTrust-Aware Enterprise Credit Risk Prediction via Variational Autoencoder in Supply Chain Finance
The growing complexity of transaction networks, dynamic enterprise interactions, and rising financial uncertainties in modern supply chains have made accurate credit risk assessment for small and medium enterprises (SMEs) in supply chain finance (SCF) a critical priority. However, traditional credit prediction models struggle to capture meaningful latent risk patterns and dynamic risk propagation within supply chain networks, limiting their effectiveness, robustness, and interpretability in supporting trustworthy decision-making in SCF ecosystems. To this end, we propose TrustVAE , a novel end-to-end generative paradigm that jointly models iterative node aggregation, trust-aware message passing, and spatio-temporal dynamics for credit risk assessment in SCF. Specifically, TrustVAE leverages the Iterative Latent Risk Aggregation (ILRA) module to extract latent risk representations from raw enterprise data in an unsupervised manner through deep nonlinear transformations and iterative clustering, capturing inherent structural properties that extend beyond the static financial indicators used in conventional models. Meanwhile, the Bounded Graph Attention Network (BGAN) module integrates trust-aware bounded confidence into graph-based risk propagation, enhancing the fidelity of diffusion modeling by suppressing risk transmission across low-trust enterprise edges. Furthermore, TrustVAE introduces the Temporal Graph Autoregressive Flow (TGAF) module, which couples temporal dependencies with autoregressive flow modeling to characterize credit risk evolution across multi-scale temporal horizons (e.g., daily, weekly, and monthly), jointly capturing dynamic patterns in both temporal and graph-structured domains. Extensive experiments on real-world datasets demonstrate that TrustVAE outperforms existing competitors, achieving average absolute improvements of 1.59%, 3.17%, 1.93%, and 2.11% in ROC-AUC, AR, Accuracy, and F1 Score, respectively. The unified design of TrustVAE strengthens the practical foundation of information systems in addressing credit risk within data-intensive, networked environments, offering novel insights into generative modeling for credit risk prediction.
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (9) , 5854-5866
Evolution Rather Than Degradation: Structure-Guided Elastic Consensus Learning for Multimodal Knowledge Graph Completion
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (9) , 5854-5866
Evolution Rather Than Degradation: Structure-Guided Elastic Consensus Learning for Multimodal Knowledge Graph Completion
- 18 June 2026
- Published by Elsevier in Pattern Recognition
- Vol. 180
Evolution-consistent dynamic graph condensation
- 18 June 2026
- Published by Elsevier in Pattern Recognition
- Vol. 180
Evolution-consistent dynamic graph condensation
- 10 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (8) , 5232-5244
Knowledge Graph-Based Debiasing for Trustworthy Recommendation Systems
These years have witnessed remarkable progress in modeling user behaviour from personalized online services, especially knowledge graph-based recommendation systems. Meanwhile, more studies are focusing on aspects beyond recommendation performance, since such an observational data-driven paradigm is posing threats to both users and society in terms of trustworthiness. In fact, existing problem-oriented solutions still face significant challenges, as almost all of them suffer from the generality limitations to improve their trustworthiness in a uniform fashion. To address these issues, we propose a plug-and-play Debiasing framework for Knowledge Graph-based Recommendation Systems, also known as DiKGRS. Specifically, the Knowledge-augmented Pseudo-Samples Generation (KPSG) method, a novel data augmentation perspective, is proposed to explore more auxiliary information beyond observational user behaviors. Furthermore, the Debiasing Value Networks (DVN), is also developed to evaluate the reliability of generated pseudo-samples by modeling both the item popularity and user demographic bias in the platform. Moreover, an adaptive weighting coordination module is performed to coordinate the proposed DiKGRS framework and its backbones. Experimental results on four real-world datasets from different online service personalization scenarios have illustrated that the proposed framework can significantly improve the trustworthiness of existing knowledge graph-based recommendation systems. The code has been released public available at: https://github.com/alipay/A-Knowledge-augmented-Method-DiKGRS.
- 10 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (8) , 5232-5244
Knowledge Graph-Based Debiasing for Trustworthy Recommendation Systems
These years have witnessed remarkable progress in modeling user behaviour from personalized online services, especially knowledge graph-based recommendation systems. Meanwhile, more studies are focusing on aspects beyond recommendation performance, since such an observational data-driven paradigm is posing threats to both users and society in terms of trustworthiness. In fact, existing problem-oriented solutions still face significant challenges, as almost all of them suffer from the generality limitations to improve their trustworthiness in a uniform fashion. To address these issues, we propose a plug-and-play Debiasing framework for Knowledge Graph-based Recommendation Systems, also known as DiKGRS. Specifically, the Knowledge-augmented Pseudo-Samples Generation (KPSG) method, a novel data augmentation perspective, is proposed to explore more auxiliary information beyond observational user behaviors. Furthermore, the Debiasing Value Networks (DVN), is also developed to evaluate the reliability of generated pseudo-samples by modeling both the item popularity and user demographic bias in the platform. Moreover, an adaptive weighting coordination module is performed to coordinate the proposed DiKGRS framework and its backbones. Experimental results on four real-world datasets from different online service personalization scenarios have illustrated that the proposed framework can significantly improve the trustworthiness of existing knowledge graph-based recommendation systems. The code has been released public available at: https://github.com/alipay/A-Knowledge-augmented-Method-DiKGRS.
- 4 June 2026
- Published by Elsevier in Pattern Recognition
- Vol. 180
Multi-level decoupled trend learning for GNN-based multivariate time series prediction
- 4 June 2026
- Published by Elsevier in Pattern Recognition
- Vol. 180
Multi-level decoupled trend learning for GNN-based multivariate time series prediction
- 27 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (4) , 2444-2456
Spatial-Temporal Multimodal Large Language Model for Generative Recommendation in Alipay
Despite the encouraging achievements, the practical application of recommendation systems still faces two key issues. The first is how to better understand the multimodal real-time requests that are the more mainstream request behavior in industrial scenarios; the other is how to effectively capture users’ dynamic needs that change with temporal and spatial conditions. The breakthroughs in text understanding and generation capabilities of Large Language Models (LLMs) have demonstrated their tremendous potential in precise recommendation systems, particularly through the enhancement of the understanding of user intent. To address these issues, we propose a novel Spatial-Temporal Multimodal LLM for generative recommendation. Specifically, on the basis of the behavior data constructed from Alipay, spatial-temporal knowledge-guided fine-tuning module is proposed to capture specific needs in user real-time requests. Furthermore, a preference discovery module is developed to learn user preferences in visual queries from multimodal request perspective. Meanwhile, a personalized recommendation module is designed to aggregate spatial-temporal knowledge and user preferences for generative recommendation. Experimental results on a real-world deployed generative recommendation task from the ‘Explore’ scenario in Alipay have demonstrated the effectiveness of the proposed framework.
- 27 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Knowledge and Data Engineering
- Vol. 38 (4) , 2444-2456
Spatial-Temporal Multimodal Large Language Model for Generative Recommendation in Alipay
Despite the encouraging achievements, the practical application of recommendation systems still faces two key issues. The first is how to better understand the multimodal real-time requests that are the more mainstream request behavior in industrial scenarios; the other is how to effectively capture users’ dynamic needs that change with temporal and spatial conditions. The breakthroughs in text understanding and generation capabilities of Large Language Models (LLMs) have demonstrated their tremendous potential in precise recommendation systems, particularly through the enhancement of the understanding of user intent. To address these issues, we propose a novel Spatial-Temporal Multimodal LLM for generative recommendation. Specifically, on the basis of the behavior data constructed from Alipay, spatial-temporal knowledge-guided fine-tuning module is proposed to capture specific needs in user real-time requests. Furthermore, a preference discovery module is developed to learn user preferences in visual queries from multimodal request perspective. Meanwhile, a personalized recommendation module is designed to aggregate spatial-temporal knowledge and user preferences for generative recommendation. Experimental results on a real-world deployed generative recommendation task from the ‘Explore’ scenario in Alipay have demonstrated the effectiveness of the proposed framework.
- 6 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 37 (7) , 3349-3359
HarmoFGL: Harmonizing GNN Latent Factors for Federated Graph Learning
Federated graph learning (FGL), as a privacy-preserving paradigm for distributed graph data training, aims to resolve graph data isolation issues under the framework of federated learning (FL). Despite the significant efforts made by existing FGL methods, two key challenges are still not well addressed: 1) how to mitigate graph heterogeneity in clients arising from feature deviation and structural deviation and 2) how to devise a favorable aggregation mechanism to maximize the client’s benefit from collaborative training with privacy preserving. To tackle these issues, we take a perspective of latent factor and propose a HarmoFGL framework by Harmonizing graph neural network (GNN) latent factors for Federated Graph Learning, achieving cross-client federated training by coordinating personalized aggregation and client-level representation in a symbiotic space. To alleviate feature deviation, an implicit feature crossing (IFC) approach is proposed through the disentanglement of higher order feature dependency into client-universal and client-specific interactions. As for the graph heterogeneity induced by structural deviation, we establish a cross-client symbiotic parameter space spanned by GNN latent factors, on which a client-level representation is derived to characterize the inherent properties of clients. On the server side, on the basis of client relevance-driven personalized parameter aggregation, graph Laplacian regularization on client-level representations is implemented for collaborative training. Experimental results on five public graph datasets and two medical datasets demonstrate the effectiveness of HarmoFGL.
- 6 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 37 (7) , 3349-3359
HarmoFGL: Harmonizing GNN Latent Factors for Federated Graph Learning
Federated graph learning (FGL), as a privacy-preserving paradigm for distributed graph data training, aims to resolve graph data isolation issues under the framework of federated learning (FL). Despite the significant efforts made by existing FGL methods, two key challenges are still not well addressed: 1) how to mitigate graph heterogeneity in clients arising from feature deviation and structural deviation and 2) how to devise a favorable aggregation mechanism to maximize the client’s benefit from collaborative training with privacy preserving. To tackle these issues, we take a perspective of latent factor and propose a HarmoFGL framework by Harmonizing graph neural network (GNN) latent factors for Federated Graph Learning, achieving cross-client federated training by coordinating personalized aggregation and client-level representation in a symbiotic space. To alleviate feature deviation, an implicit feature crossing (IFC) approach is proposed through the disentanglement of higher order feature dependency into client-universal and client-specific interactions. As for the graph heterogeneity induced by structural deviation, we establish a cross-client symbiotic parameter space spanned by GNN latent factors, on which a client-level representation is derived to characterize the inherent properties of clients. On the server side, on the basis of client relevance-driven personalized parameter aggregation, graph Laplacian regularization on client-level representations is implemented for collaborative training. Experimental results on five public graph datasets and two medical datasets demonstrate the effectiveness of HarmoFGL.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (9) , 1-15
MRCNet: Motion Reasoning Chain for Cross Modal Video Camouflaged Object Detection
Video camouflaged object detection (VCOD) aims to identify objects that seamlessly blend into their surroundings in video sequences. Traditional methods merely rely on visual cues to capture inter-frame motion that reveals camouflaged objects. However, the high similarity between camouflaged objects and their environments often renders pure reliance on visual cues unreliable. Additionally, random motions including camera shaking and abrupt scene transitions also inevitably bring noise into the identification process. To overcome these challenges, we propose a Motion Reasoning Chain Network (MRCNet), a novel cross-modal VCOD framework that emulates the human thought process when observing camouflaged objects, i.e., motion reasoning. Specifically, we introduce a generative sampling strategy grounded in multimodal large language models (MLLMs) to bridge the implicit knowledge space of MLLMs and the explicit representation space regarding the attributes of camouflaged objects, thereby enabling the effective establishment of the motion reasoning chain tailored for VCOD. This process provides semantic guidance for visual comprehension of camouflaged objects through motion and concept attribute reasoning. To improve the identification capability of camouflaged objects, we develop motion representation learning driven by the motion reasoning chain. It introduces hierarchical de-biased motion prototype learning to mitigate hallucinations of MLLMs, boosting the motion perception. To learn precise prompts for the visual foundation model, cross-modal prompt learning further incorporates the de-biased concept prototype into visual representations to enhance the visual comprehension of camouflaged objects. Extensive experiments across three datasets demonstrate that MRCNet achieves state-of-the-art results on both general metrics and spatiotemporal consistency metrics.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (9) , 1-15
MRCNet: Motion Reasoning Chain for Cross Modal Video Camouflaged Object Detection
Video camouflaged object detection (VCOD) aims to identify objects that seamlessly blend into their surroundings in video sequences. Traditional methods merely rely on visual cues to capture inter-frame motion that reveals camouflaged objects. However, the high similarity between camouflaged objects and their environments often renders pure reliance on visual cues unreliable. Additionally, random motions including camera shaking and abrupt scene transitions also inevitably bring noise into the identification process. To overcome these challenges, we propose a Motion Reasoning Chain Network (MRCNet), a novel cross-modal VCOD framework that emulates the human thought process when observing camouflaged objects, i.e., motion reasoning. Specifically, we introduce a generative sampling strategy grounded in multimodal large language models (MLLMs) to bridge the implicit knowledge space of MLLMs and the explicit representation space regarding the attributes of camouflaged objects, thereby enabling the effective establishment of the motion reasoning chain tailored for VCOD. This process provides semantic guidance for visual comprehension of camouflaged objects through motion and concept attribute reasoning. To improve the identification capability of camouflaged objects, we develop motion representation learning driven by the motion reasoning chain. It introduces hierarchical de-biased motion prototype learning to mitigate hallucinations of MLLMs, boosting the motion perception. To learn precise prompts for the visual foundation model, cross-modal prompt learning further incorporates the de-biased concept prototype into visual representations to enhance the visual comprehension of camouflaged objects. Extensive experiments across three datasets demonstrate that MRCNet achieves state-of-the-art results on both general metrics and spatiotemporal consistency metrics.